Professional Documents
Culture Documents
Article 7
Article 7
North-Holland
Scott L. LUMMER
Te.rus A & M Universi<v. College Stution. TX 77843. USA
John J. MCCONNELL
Purdue Unioersi{v. War Lu/uyette. IN 47907. ti.SA
This paper investigates the hypothesis that bank loans convey information to the capital market ’
regarding the value of the borrowing firm. Unlike previous researchers. we distinguish between
new bank loans and loan renewals. For new loans, the excess stock return for borrowers around
the loan announcement is not significantly different from zero. For favorable loan revisions, the
excess return is significantly positive: for unfavorable revisions. it is significantly negative. We
interpret these results to imply that banks play an important role as transmitters of information in
capital markets. but new bank loans per se do not communicate information.
1. Introduction
*This paper has benefited from presentations at the University of North Carolina, the Gam
Institute of Finance. the Texas Finance Symposium, and the Center for Research in Security
Prices at the University of Chicago: the comments of David Dubofsky. Mark Flannety, Donald
Fraser, James Kolari. William Kracaw, Wanda Wallace, Christopher James (the referee), and Cliff
Smith (the editor): and the research assistance of Horace Ho.
There are at least two perspectives on the way in which banks gain access to
information not available to other capital-market participants. According to
‘See Asquith and Mullins (1986). Dann and Mikkelson (1984). Eckbo (1986). Linn and Pinepar
(1988). Loderer and Van Drunen (1986). and Masulis and Korwar (1986). Smith (1986) summa-
rizes this literature. Two exceptions to this general observation are Kim and Stulz (1988). who
report a positive and significant excess return around the announcement of Eurobond issues. and
Wruck (1989). who documents a positive and significant excess return around the announcement
of private equity offerings.
the first. banks invest in information-gathering technology that gives them a
competitive advantage in evaluating risky lending opportunities. When a
potential borrower applies for a loan, the bank evaluates the borrower, and the
bank’s loan decision signals the prospective borrower’s creditworthiness to
other capital-market participants. Benston and Smith (1976), Diamond (1984),
and Campbell and Kracaw (1980), among others. develop this idea more fully.
If it is assumed that a firm will enter into a new bank loan agreement only if it
currently has no bank financing in place or the terms of the new credit
agreement are more favorable than its current agreement. this line of reasoning
predicts a positive stock-price response when new bank loans are announced.
An alternative view is that banks gain access to private information about
their customers over time as a result of an intimate. continuing business
relationship with them. This idea can be traced to Black (1975) and Kane and
Malkiel (1965). Fama (1985) expands on this theme to argue that banks play a
unique role in providing funds to businesses. Fama’s argument is composed of
two parts. First, bank debt, along with other types of privately placed
fixed-payoff securities, is classified as inside debt. Banks have access to
information not available to holders of the firm’s publicly traded securities or
those who hold other outside claims, such as employees and trade creditors.
Second, because bank loans typically have a low priority among fixed-payoff
claims, signals from the credit renewal process are credible and consequently
reduce the monitoring costs incurred by the firm’s other claimants.
Fama’s argument for the uniqueness of bank loans places considerable
weight on the loan renewal process as a mechanism for transmitting informa-
tion. Loan renewals are important because of the periodic review to which
firms that enter into short-term bank credit agreements submit themselves.
Within this framework, there is no requirement that banks have a competitive
information advantage over other suppliers of funds at the initiation of the
loan. Rather, banks learn about their customers through time as a natural
outgrowth of their business interactions. This line of reasoning suggests that if
there is to be a stock-price response to announcements of bank credit
agreements, the effect should be observed around announcements of revisions
to, not initiations of, such agreements.’ Of course, announcements regarding
loan revisions or renewals can signal either positive or negative information.
For example, a revision in which the interest rate on the loan is reduced or
restrictive convenants are relaxed is likely to provide positive information
about the state of the firm. Alternatively, an announcement in which the credit
‘A variation on this theme is that a firm’s willingness to submit to periodic credit evaluations
may provide market participants with a positive signal concerning management’s assessment of
the firm’s prospects. If so. announcements of new credit agreements should be associated with a
positive stock-price reaction even if banks have no inside information at the initiation of the credit
agreement. In order for this scenario to lead to a separating equilibrium. firms must face a penalty
for false signaling.
102 S. L. Lumnwr und J.J. .McConnell. Bank lounr us slgnuls ojfirm due
Table 1
Frequency distribution by year for a sample of 371 new bank credit agreements and 357 revised
bank credit agreements for NYSE- and AMEX-listed companies. 1976-1986.
Credit agreements
Credit agreements with new banks
All credit All new credit denoted as new identified in Revised credit
Year agreements agreements in annual report annual report agreements
1976 X6 37 23 8 49
1977 103 44 25 8 59
1978 94 52 33 15 42
1979 105 62 36 11 . 43
1980 83 49 23 7 34
1981 42 26 9 5 16
1982 66 36 23 8 30
1983 47 17 11 30
1984 47 21 15 : 26
1985 33 18 9 4 15
1986 22 9 7 2 13
- - -
Total 728 371 214 58 357
104 S. L Lumnter und J.J. McConnell, Bunk louns 11ss~gnuls of,th due
Table 2
Descriptive statistics for a sample of 371 new bank credit agreements and 357 revised bank credit
agreements for NYSE- and AMEX-fisted companies. 1976-1986.
aNumber of shares of common stock outstanding multiplied by the market price per share five
days before the announcement of the credit agreement.
When the information is available, the dollar amount of the loan and the
term-to-expiration of the credit agreement are recorded. Table 2 displays
summary statistics for these data.
4. Methodology
5. I. Full-sample results
Table 3
Average announcement-period excess returns, significance tests. and proportion of positive excess
returns for a sample of 371 new bank credit agreements and 357 revised bank credit agreements
for NYSE- and AMEX-listed companies, 1976-1986.
Announcement-
Announcement- period proportion
Number of period excess z- of positive
Type of announcement observations return (%) statistic excess returns
‘For the observations in this subsample, some terms of the new credit agreement are more
favorable than those in the old agreement and some terms are less favorable.
bNo negative news concerning the credit agreement was published in the WSJ in the twelve
months prior to the revision.
‘Negative news concerning the credit agreement was published in the WSJ within twelve
months prior to the revision.
dAnnouncement-period excess return is significantly different from zero at the 0.01 level.
‘Proportion of positive announcement-period excess returns is significantly different from the
pryportion of positive residuals during the market-model estimation period at the 0.01 level.
Proportion of positive announcement-period excess returns is significantly different from the
proportion of positive residuals during the market-model estimation period at the 0.05 level.
favorably, (2) the terms are revised unfavorably, or (3) some terms are revised
favorably, while others are revised unfavorably.
There are four dimensions of a credit agreement by which the terms of a
loan can be revised: its maturity, interest rate, dollar value. and protective
covenants. The protective covenants include such items as a minimum current
ratio, a maximum leverage ratio, and the security pledged against the loan.
Observations are placed in the favorably-revised category if the WSJ article
indicates that the maturity of the agreement is lengthened. the interest rate is
reduced, the dollar value of the loan is increased, or the protective covenants
are made less restrictive. In 43 cases the WSJ article reports that the loan
‘replaced’ an existing credit agreement, but gives no specific information about
the terms of the previous loan. In those cases, we assume that the only
provision that changes is the maturity date, so those observations are placed
into the favorably-revised category. This category contains 259 observations.
The Skill Corp. and Genisco Technology Corp. announcements quoted earlier
are examples of favorable revisions.
The second category contains observations in which the agreement is revised
unfavorably. Either the amount of the loan is reduced, the maturity is
shortened, the interest rate is increased, or the protective convenants are made
more severe. This group contains 22 observations. One way to simultaneously
reduce the amount of the loan and decrease its maturity is to cancel the credit
agreement. This occurs in eight cases. The following announcement concerning
Storage Technology Corp. is an example of an unfavorable revision to a credit
agreement:
To a limited extent, the results are consistent with the idea that more
positive information is revealed when the bank signals its willingness to
continue to work with a firm when it has been previously reported that there is
a problem with the loan. For this sample, the announcement-period excess
return is + 4.82%, which, with a z-statistic of + 4.08, is significantly positive.
For the other 26 observations, the average two-day excess return of + 2.35% is
not significantly different from zero. These results suggest that the market
reacts more strongly when the bank signals its intent to continue to work with
a client known to be in distress. This type of information signal could be
important in helping a struggling firm to continue operations because of its
effect on other parties doing business with the firm. However, the average
excess returns for the two groups are not significantly different from each
other. Of course, in conducting this test, we are relying heavily on information
provided by the WSJ. In some cases, the security market may have informa-
tion that a credit agreement is in distress despite the lack of such a report in
the WSJ. To the extent we have misclassified these announcements, the
announcement-period excess return for this sample is biased upward and’the’
power of our test reduced.
Except for five cases, all of the firms in our sample that announce new credit
agreements had some prior bank financing in place, albeit with a different
bank. Thus, just as with loan revisions, the new bank loan can be made on
terms that are more or less favorable than those of the old credit agreement.
To classify the terms of each new credit agreement as more or less favorable,
112 S. L. Lumnwr und J.J. .McConnell, Bunk loons us rrgnuls of’firm wiur
we searched the company’s annual reports to identify the terms of both the old
and the new agreement. We are able to identify both sets of terms for 198 of
the 371 new-loan announcements. Among this group, 180 new loans have
more favorable terms than the old agreement and 18 have some terms that are
more favorable and some that are less favorable. We are able to identify no
cases in which all the terms are less favorable than the old agreement. Of the
180 new loans on more favorable terms, 135 are from the set identified as a
new loan in the annual report and 58 are from the set for which the annual
Table 4
Average announcement-period excess returns. significance tests, and proportion of positive excess
returns for a sample of 198 new bank credit agreements for which the terms of the new and old
agreements could be determined for NYSE- and AMEX-listed companies. 1976-1986.
Announcement-
Announcement- period proportion
Number of period excess :- of positive
Type of announcement observations return (%jd statistic excess return?
(A) /Vew credrr ugreemetlts made LWmorefworuhle terms thon prior ugreement
(B) New credit ugreements with II nuxture of more und less furwuhle terms
reiutrre to prior credit agreementa
(C) New credit agreements with (I mixture o/more and less fworuhle terms
relutiL)e to prior credit ugreement
aFor observations in this subsample, some terms of the new credit agreement are more
favorable than those of the old agreement and some terms are less favorable than those of the old
agreement.
b No negative news concerning the old credit agreement was published in the WSJ in the twelve
months prior to the revision.
‘Negative news concerning the old credit agreement was published in the WSJ in the twelve
months prior to the revision.
d No announcement-period excess return is significantly different from zero at the 0.05 level.
eNo proportion of positive announcement-period excess returns is significantly different from
the proportion of positive residuals during the market-model estimation period at the 0.05 level.
report indicates that the new agreement involves a new bank. A similar
breakdown of the new loans with mixed terms is 12 and 5. respectively. To
parallel loan revisions with mixed terms. new loans with mixed terms are
classified according to whether the IKSJ previously has reported that the
borrower was in danger of violating one or more of the covenants of the old
credit agreement.
Excess returns for the various categories of new credit agreements are
presented in table 4. They are easily summarized: in no case is the announce-
ment-period excess return significantly different from zero. Additionally, for
each of the three categories of more favorable new loans (panel A), the excess
return is less than the excess return for the sample of favorably revised loans.
It is significantly less (at the 0.05 level) for the two largest samples. For each of
the three categories of mixed-term new loans (panel B), the excess return is
less than the excess return for the sample of mixed revisions. It is significantly
less for the two largest samples.
The evidence indicates that the bank lending process works in the following
way: When a bank enters into a new credit agreement, it does so with no
consequential information advantage over other outside claimholders and, on
average, announcements of new loan agreements reveal no information. even if
the new loan is on more favorable terms than the firm’s old loan. Over time,
the bank becomes privy to information not available to outside claimholders,
and, based on this information, periodically revises the terms of the credit
agreement. If the information available to the bank reflects positively on the
firm, the loan can be renewed or revised on terms more favorable to the
borrower. This decision sends a positive signal to the market. Alternatively, if
the firm is having financial difficulties. the bank can cancel the loan. increase
the interest rate, or tighten various protective covenants. This decision signals
negative information to the market.
There is a third course of action the bank can take. If the firm is having
trouble meeting a particular loan covenant, the bank can restructure the loan
to permit the firm to continue operations. The strength of the signal provided
by this decision depends on what information was previously available to the
market. If the market is already aware of the problem, the stock-price reaction
is more positive than if the problem is first revealed publicly by the loan
restructuring. In short, the data indicate that the bank loan review and
revision process provides useful information to capital-market participants.
The results are not totally satisfyin,. 0 however. If new loan announcements
reveal no information and if information is revealed only as credit agreements
are revised, the average announcement-period excess return across all types of
114 S. L. Lummer and J.J. .McConnell. Rank loons us signuls offirmdue
loan revisions should be zero. That is, on average, excess returns around
announcements in which credit agreements are revised on more favorable
terms should just offset those in which loans are cancelled or revised unfavor-
ably. Contrarily, across all credit agreement revisions, the average announce-
ment-period excess return is positive and statistically significant. One possible
explanation for this phenomenon is that there is a reporting bias on the part of
the firms or banks. They may be less inclined to announce that performance
has been unsatisfactory and that credit agreements have been terminated. An
alternative explanation is that in many cases a credit agreement may simply be
allowed to expire. Such expirations can represent a negative decision by the
bank, but as with dividend omissions, an announcement is not required and
our data collection procedure cannot identify such occurrences. The sheer
difference in the sizes of our samples lends some support to this conjecture.
The sample contains 259 favorably-revised agreements and only 22 that are
revised unfavorably.
Bias could also creep into the sample through our screening process. By
including only clean announcements, we may have screened out a greater
proportion of negative than positive revisions. To investigate the possible
impact of such a bias, we generate the announcement-period excess return for
the entire sample of 456 clean and contaminated announcements of credit
agreement revisions for which return data are available. For this sample, the
average announcement-period excess return is -t-0.81%, which, with a z-statis-
tic of + 2.88, is significantly different from zero. Thus, the announcement-
period return is smaller when clean and contaminated announcements are
used, but it is still significantly positive.
Table 5
Descriptive statistics for a sample of 180 new bank credit agreements made on more favorable
terms than the previous agreements and 259 favorably-revised bank credit agreements for NYSE-
and AMEX-listed companies. 1976-1986.
Favorably-revised
New agreements agreements
Variable Range Mean Median Range Mean Median
‘Number of shares of common stock outstanding multiplied by the market price per share five
days prior to announcement of the credit agreement.
favorably-revised credit agreements and those with new loans made on more
favorable terms than the previous credit agreement. Descriptive statistics for
these two samples are presented in table 5. The differences between the two
groups are far less dramatic than those shown in table 2. Still, it is possible
that the loans and the firms in the sample of favorably-revised loans differ
from those in the sample of new loans, and it’ is this difference that drives the
differences in the announcement-period excess returns for the two groups.
Loans can differ on a number of dimensions. We consider four on which we
have data for at least some of the credit agreements: (1) relative size, (2)
maturity, (3) whether the loans are secured or unsecured, and (4) structure
(whether the loan is a revolving credit agreement or a term loan). The various
samples of favorably-revised credit agreements and new loans made on more
favorable terms are categorized on these dimensions and announcement-period
returns are generated. The results, presented in table 6, panel A through D, do
not suggest that the differential in excess returns between new and revised
loans is due to any of these characteristics. For the sample of favorably-revised
loans, the announcement-period excess return is positive for 10 of the 11
subgroups of loans considered and it significantly greater than zero for 6 of
them. Contrarily, for the various samples of new credit agreements, approxi-
mately half of the announcement-period excess returns are negative, half are
positive, and none is significantly different from zero. Thus, on the basis of
univariate tests, the distinguishing characteristic among bank credit agree-
ments - at least so far as the capital-market response is concerned - is whether
the agreement is new or revised.
116 S. L. Lummer and J.J. ,McCo~meil. Bunh louns us srgnnk ojfirm wlue
Table 6
Average announcement-period excess returns, signiticance tests. and proportion of positive excess
returns for a sample of 180 new bank credit agreements made on more favorable terms than
previous agreements and 259 favorably-revised bank credit agreements for ?;YSE- and AMEX-
listed companies with subsamples based on dollar value, maturity. security. and structure of the
credit agreement.
Announcement-
Announcement- period proportion
Number of period excess z- of positive
Sample observations return(F) statistic excess returns
(C) Clussificution of sumpies bused ott security supporting the credit ug;eememb
All new credit agreements
Unsecured 1x - 0.29 -0.93 0.444
Secured 13 - 0.57 0.42 0.538
Security unknown 148 0.15 1.14 0.459
S. L. Lummer und J.J. McCormell. Bank loans us stgnals ojjrm due 117
Table 6 (continued)
Announcement-
Announcement- period proportion
Number of period excess z- of positive
Sample observations return (c?c) statistic excess returns
‘Relative value is the dollar value of the agreement divided by the market value of the firm’s
common equity five days before to the announcement. The amount of the agreement is unknown
for one new agreement and two favorably-revised agreements.
bOne new agreement and one favorably-revised agreement were guaranteed by a government or
government agency. Those observations are not included in any of these subsamples.
‘Announcement-period excess return is significantly different from zero at the 0.01 level.
dAnnouncement-period excess return is significantly different from zero at the 0.05 level.
eProportion of positive announcement-period excess returns is significantly different from the
pryportion of positive residuals during the market-model estimation period at the 0.05 level.
Proportion of positive announcement-period excess returns is significantly different from the
proportion of positive residuals during the market-model estimation period at the 0.01 level.
3.8. Multivariate analysis
where ERi is the two-day excess announcement return for firm i, B,, . . . , B,,
are the regression coefficients, X,, . . . , Xl, are the variables described above,
and ei is the disturbance term with zero mean. Because cross-sectional stock
returns exhibit heteroskedaticity, both sides of the regression equation are
divided by si, where s, is the standard deviation of the prediction derived from
the market-model estimation. The revised regression equation is
Table 7
Results of regression of standarized excess returns on various standarized continuous and dummy
variables for sample of 180 new bank credit agreements made on more favorable terms than the
previous credit agreement and 259 favorably-revised bank credit agreements for NYSE- and
AMEX-listed companies, 1976-1986 (r-statistics in parentheses).
aComputed as log of the maturity of the credit agreement divided by estimated standard error
ofbtwo-day return.
Computed as amount of the credit agreement divided by the market value of equity. and
divided by estimated standard error of two-day return.
‘Coefficient is significantly different from zero at the 0.10 level.
where SER, is the standarized announcement-period excess return for an-
nouncement i.
The regression results are presented in table 7. The coefficient of the dummy
variable indicating whether the loan is a revision or new is positive in each
regression, but is significantly greater than zero at only the 0.09, 0.04, and 0.13
levels. respectively, for the three samples. Thus, the results are consistent with
(albeit at weak levels of significance) the hypothesis that announcements of
favorably-revised bank credit agreements provide a positive signal to outside
claimholders and that announcements of new bank loans do not.
References
Asquith. Paul and David .Mullins, 1986, Equity issues and offering dilution. Journal of Financial
Economics 15. 61-90.
Benston, George and Clifford Smith, 1976. A transactions cost approach to the theory of financial
intermediation. Journal of Finance 31, 215-231.
Black. Fischer. 1975. Bank funds management in an efficient market, Journal of Financial
Economics 2. 323-339.
Brown. Stephen and Jerold Warner, 1985. Using daily stock returns: The case of event studies.
Journal of Financial Economics 14, 3-31.
Campbell. Tim and William Kracaw. 1980, Information production. market signaling. and the
theory of intermediation, Journal of Finance 35. 863-882.
Dann. Larry and Wayne Mikkelson. 1984, Convertible debt issuance, capital structure change and
financing-related information: Some new evidence, Journal of Financial Economics 13.
157-186.
Diamond. Douglas, 1984. Financial intermediation and delegated monitoring, Review of Eco-
nomic Studies 51. 393-414.
Echbo. Espen, 198% Valuation effects of corporate debt offerings, Journal of Financial Economics
15,119-151.
Fama. Eugene. 1985. What’s different about banks?, Journal of Monetary Economics 15. 29-39.
Holthausen. Robert and Richard Leftwich. 1986. The effect of bond rating changes on common
stock prices. Journal of Financial Economics 17, 57-89.
122 S. L. Lummer and J.J. McConneii. Bank loans us signals offirm tulue
James, Christopher, 1987. Some evidence on the uniqueness of bank loans, Journal of Financial
Economics 19. 217-235.
Kane, Edward and Burton Malkiel. 1965, Bank portfolio allocation. deposit variability and the
availability doctrine, Quarterly Journal of Economics 79. 113-134.
Kim, Yong and Rene Stulz. 1988, The Eurobond market and corporate financial policy: A test of
the clientele hypothesis, Journal of Financial Economics 22, 189-206.
Linn. Scott and Michael Pinegar, 1988, The effect of issuing preferred stock on common and
preferred stockholder wealth, Journal of Financial Economics 22, 155-184.
Loderer. Claudio and Leonard Van Dnmen, 1988, The price elasticity of demand for common
stock: An empirical investigation, Unpublished manuscript (Purdue University, West Lafayette.
IN).
Masulis, Ronald and Ashok Kotwar, 1986, Seasoned equity offerings: An empirical investigation.
Journal of Financial Economics 15, 91-118.
Mikkelson. Wayne and Megan Partch, 1986, Valuation effects of security offerings and the
issuance process, Journal of Financial Economics 15, 31-60.
Smith. Clifford. 1986. Investment banking and the capital acquisition process. Journal of Finan-
cial Economics 15. 3-29.
Wruck. Karen. 1989. Equity ownership concentration and firm value: Evidence from private
equity tinancings. Journal of Financial Economics 23, 3-28.